Analysis of Immigrants in U.S
Prateek Rajput
18/12/2019
Data Summary: The data used in this project is from 2017 American Community Survey(ACS). Two types of data was made available, Household data and Population data. The one which I chose is the Population data. I selected 19 columns, but all of them were not used in my project. The one’s that felt useful to me were:
ST: State code.
AGEP: Age of a person.
CIT: Citizenship status.
COW: Class of worker.
NWLK: Looking for work.
SCHL: Educational Attainment.
SCHG: Grade Level Attending.
SEX: Gender of a person.
WKL: When last worked.
ESR: Employment status recode.
FOD1P: Field of degree.
PINCP: Total person’s income.
POBP: Place of birth of a person.
RAC3P: Recorded detailed race code.
ADJINC: Adjustment factor for income.
The main aim of this project is to glance over the numbers about immigrants in the States of U.S.A. Therefore, I did and exploratory analysis of immigrants in the U.S.
Pre Processing of the data: I created a string of countries’ code and name to use it later in showing some of the data of the countries on world map.
person_1<- read_csv("psam_pusa.csv",col_types = cols_only(RT=col_character(),SERIALNO=col_character(),AGEP=col_integer(),CIT=col_character(),ST=col_character(),COW=col_character(),NWAB=col_character(),NWAV=col_character(),NWLA=col_character(),NWLK=col_character(),SEX=col_character(),WKL=col_character(),ANC1P=col_character(),ANC2P=col_character(),FOD1P=col_character(),FOD2P=col_character(),NATIVITY=col_character(),POBP=col_character(),RAC3P=col_character(),SCHL=col_character(),ESR=col_character(),YOEP=col_character(),PERNP=col_character()))
person_2<- read_csv("psam_pusb.csv",col_types = cols_only(RT=col_character(),SERIALNO=col_character(),AGEP=col_integer(),CIT=col_character(),ST=col_character(),COW=col_character(),NWAB=col_character(),NWAV=col_character(),NWLA=col_character(),NWLK=col_character(),SEX=col_character(),WKL=col_character(),ANC1P=col_character(),ANC2P=col_character(),FOD1P=col_character(),FOD2P=col_character(),NATIVITY=col_character(),POBP=col_character(),RAC3P=col_character(),SCHL=col_character(),ESR=col_character(),YOEP=col_character(),PERNP=col_character()))
person_3<- read_csv("psam_pusc.csv",col_types = cols_only(RT=col_character(),SERIALNO=col_character(),AGEP=col_integer(),CIT=col_character(),ST=col_character(),COW=col_character(),NWAB=col_character(),NWAV=col_character(),NWLA=col_character(),NWLK=col_character(),SEX=col_character(),WKL=col_character(),ANC1P=col_character(),ANC2P=col_character(),FOD1P=col_character(),FOD2P=col_character(),NATIVITY=col_character(),POBP=col_character(),RAC3P=col_character(),SCHL=col_character(),ESR=col_character(),YOEP=col_character(),PERNP=col_character()))
person_4<- read_csv("psam_pusd.csv",col_types = cols_only(RT=col_character(),SERIALNO=col_character(),AGEP=col_integer(),CIT=col_character(),ST=col_character(),COW=col_character(),NWAB=col_character(),NWAV=col_character(),NWLA=col_character(),NWLK=col_character(),SEX=col_character(),WKL=col_character(),ANC1P=col_character(),ANC2P=col_character(),FOD1P=col_character(),FOD2P=col_character(),NATIVITY=col_character(),POBP=col_character(),RAC3P=col_character(),SCHL=col_character(),ESR=col_character(),YOEP=col_character(),PERNP=col_character()))
#binding all the 4 data frames.
full_person<- rbind(person_1,person_2,person_3,person_4)
#making temporary variables to add some new columns.
t1<- read_csv("psam_pusa.csv",col_types = cols_only(SCHG=col_guess(),ADJINC=col_guess()))
t2<- read_csv("psam_pusb.csv",col_types = cols_only(SCHG=col_guess(),ADJINC=col_guess()))
t3<- read_csv("psam_pusc.csv",col_types = cols_only(SCHG=col_guess(),ADJINC=col_guess()))
t4<- read_csv("psam_pusd.csv",col_types = cols_only(SCHG=col_guess(),ADJINC=col_guess()))
t<- rbind(t1,t2,t3,t4)
#column binding the temporary variables to the original dataframe.
full_person<- cbind(t,full_person)The full file containing all the data of person and all the columns is read in the following line. The file is in rds format.
#reading the whole file in RDS format.
full_person<- readRDS("full_person.rds")All the libraries needed:
library(readr)
library(dplyr)##
## Attaching package: 'dplyr'## The following objects are masked from 'package:stats':
##
## filter, lag## The following objects are masked from 'package:base':
##
## intersect, setdiff, setequal, unionlibrary(ggplot2)## Registered S3 methods overwritten by 'ggplot2':
## method from
## [.quosures rlang
## c.quosures rlang
## print.quosures rlanglibrary(forcats)
library(usmap)
library(tidyverse)## Registered S3 method overwritten by 'rvest':
## method from
## read_xml.response xml2## ── Attaching packages ──────────────────────────────────────────── tidyverse 1.2.1 ──## ✔ tibble 2.1.1 ✔ purrr 0.3.2
## ✔ tidyr 0.8.3 ✔ stringr 1.4.0## ── Conflicts ─────────────────────────────────────────────── tidyverse_conflicts() ──
## ✖ dplyr::filter() masks stats::filter()
## ✖ dplyr::lag() masks stats::lag()library(plotly)##
## Attaching package: 'plotly'## The following object is masked from 'package:ggplot2':
##
## last_plot## The following object is masked from 'package:stats':
##
## filter## The following object is masked from 'package:graphics':
##
## layoutlibrary(tools)
Methodology: The approcah I used to do exploratory analysis on the data of immigrants is: The first thing that I did is to find all the immigrants in the Person data. To do so, I chose all the entries with CIT==5(i.e. Not a citizen of U.S). I am assuming that all the non-citizens of the U.S are immigrants.
#non-citizens/immigrants
non_citizen <- full_person %>%
filter(CIT==5) I adjusted the income using ADJINC(adjusting factor for income).
#adjusting income.
non_citizen$PINCP<- as.integer(as.numeric(non_citizen$ADJINC) * as.numeric(non_citizen$PINCP)/1000000)I also needed the names of the states in U.S. So I installed a library called “usmap”. It has a data frame by the name “statepop” that contains the data of the states(code, abbreviation of the name, full name). I then joined the statepop and the non-citizen dataframes to add columns of the abbreviation and the full name of the states.
library(usmap)
statepop<-statepop %>% rename(ST=fips)
states_noncitizen<- inner_join(statepop,non_citizen,by="ST")
After doing all the things above,I started to explore interesting things about the data on immigrants. In this project, I mainly foused on statewise numbers of the immigrants to see if there is a difference or similarity in the immigrants in states. Therefore, I wanted to find below things:
1) Number of immigrants in each state.
2) The place of birth of the immigrants(i.e to see people from which country come to U.S).
3) Number of males and females from each country.
4) Average age of the immigrants in each state.
5) Per Capita Income of immigrants in each state.
6) If income of Male and Female immigrants were different.
7) Top Fields of degree based on income.
8) See if there are any jobless immigrants.
9) Immigrants currently pursuing bacheor’s or higher level degree(countrywise to see people from which country visit U.S to study.)
10) See if people have a bachelor’s degree but are jobless / looking for a job and see this nationwise.
11) People from which country have the highest income.
12) Check if there is a difference in the income of females and males in the Field of Degree that I will find the most earning(i.e. question 7).
13) Check if literacy affects income of graduates in a state.
14) Try to predict income using diferrent variables and also see if the number of immigrants and number of jobless immigrants affect the income of immigrants in a state.
Dealing with NAs: I removed the NAs in PINCP while doing any analysis on the income.
Dealing with income adjusted values: I adjusted the PINCP column using the ADJINC column.
Uninteresting and failed analyses:
I tried to predict the total income of immigrants in a state using total number of immigrants and jobless immigrants in that state. The r-squared value was 0.9934 which is really great, but the residual standard error was too big. I think that as the data was less(i.e 51 states), the model was overfitting the data.
Next I again tried to predict the income in a state using literacy and number of graduates in that state. I thought that more literate people earn more and the more the number of graduates in a state, more will be the total income of that state. But the r-squared was just 27%.
I also wanted to see Races of the immigrants in U.S , but I didn’t find anything interesting about that.
While exploring the data, I found that there were immigrants in armed forces of U.S, which I found very interesting, but that information was a bit irrelevant to me.
##Findings:
1) California has the most amount of immigrants.
#number of immigrants in each state.
states_df <- read.csv("https://raw.githubusercontent.com/plotly/datasets/master/2011_us_ag_exports.csv")
statewise_immigrants<- non_citizen %>%
inner_join(statepop,by="ST") %>%
group_by(abbr) %>%
summarise(immigrants=n()) %>%
arrange(desc(immigrants)) %>%
rename(state=abbr)
statewise_immigrants<-inner_join(statewise_immigrants,states_df , by=c('state'='code')) %>% select(state,immigrants)
statewise_immigrants## # A tibble: 50 x 2
## state immigrants
## <chr> <int>
## 1 CA 225964
## 2 TX 113504
## 3 NY 76376
## 4 FL 72484
## 5 NJ 35270
## 6 IL 32593
## 7 AZ 22053
## 8 GA 21944
## 9 MA 21848
## 10 WA 20448
## # … with 40 more rows#plotting the data on U.S map.
statewise_immigrants$hover <- with(statewise_immigrants, paste(state, '<br>', "Immigrants ", immigrants))
l <- list(color = toRGB("white"), width = 2)
g <- list(
scope = 'usa',
projection = list(type = 'albers usa'),
showlakes = TRUE,
lakecolor = toRGB('white')
)
plot_geo(statewise_immigrants, locationmode = 'USA-states') %>%
add_trace(
z = ~immigrants, text = ~hover, locations = ~state,
color = ~immigrants, colors = 'Purples'
) %>%
colorbar(title = "Immigrants") %>%
layout(
title = 'Number of Immigrants in each state',
geo = g
)
2) Mexicans are the one’s that migrate to U.S the most, compared to all the people from other countries.
country_df <- read.csv('https://raw.githubusercontent.com/plotly/datasets/master/2014_world_gdp_with_codes.csv') %>% rename(Country=COUNTRY)
countries <- "code,Country\n001,alabama\n002,alaska\n004,arizona\n005,arkansas\n006,california\n008,colorado\n009,connecticut\n010,delaware\n011,district of columbia\n012,florida\n013,georgia\n015,hawaii\n016,idaho\n017,illinois\n018,indiana\n019,iowa\n020,kansas\n021,kentucky\n022,louisiana\n023,maine\n024,maryland\n025,massachusetts\n026,michigan\n027,minnesota\n028,mississippi\n029,missouri\n030,montana\n031,nebraska\n032,nevada\n033,new hampshire\n034,new jersey\n035,new mexico\n036,new york\n037,north carolina\n038,north dakota\n039,ohio\n040,oklahoma\n041,oregon\n042,pennsylvania\n044,rhode island\n045,south carolina\n046,south dakota\n047,tennessee\n048,texas\n049,utah\n050,vermont\n051,virginia\n053,washington\n054,west virginia\n055,wisconsin\n056,wyoming\n060,american samoa\n066,guam\n069,commonwealth of the northern mariana islands\n072,puerto rico\n078,us virgin islands\n100,albania\n102,austria\n103,belgium\n104,bulgaria\n105,czechoslovakia\n106,denmark\n108,finland\n109,france\n110,germany\n116,greece\n117,hungary\n118,iceland\n119,ireland\n120,italy\n126,netherlands\n127,norway\n128,poland\n129,portugal\n130,azores islands\n132,romania\n134,spain\n136,sweden\n137,switzerland\n138,\"united kingdom, not specified\"\n139,england\n140,scotland\n147,yugoslavia\n148,czech republic\n149,slovakia\n150,bosnia and herzegovina\n151,croatia\n152,macedonia\n154,serbia\n156,latvia\n157,lithuania\n158,armenia\n159,azerbaijan\n160,belarus\n161,georgia\n162,moldova\n163,russia\n164,ukraine\n165,ussr\n168,montenegro\n169,\"other europe, not specified\"\n200,afghanistan\n202,bangladesh\n203,bhutan\n205,myanmar\n206,cambodia\n207,china\n208,cyprus\n209,hong kong\n210,india\n211,indonesia\n212,iran\n213,iraq\n214,israel\n215,japan\n216,jordan\n217,korea\n218,kazakhstan\n222,kuwait\n223,laos\n224,lebanon\n226,malaysia\n229,nepal\n231,pakistan\n233,philippines\n235,saudi arabia\n236,singapore\n238,sri lanka\n239,syria\n240,taiwan\n242,thailand\n243,turkey\n245,united arab emirates\n246,uzbekistan\n247,vietnam\n248,yemen\n249,asia\n253,\"south central asia, not specified\"\n254,\"other asia, not specified\"\n300,bermuda\n301,canada\n303,mexico\n310,belize\n311,costa rica\n312,el salvador\n313,guatemala\n314,honduras\n315,nicaragua\n316,panama\n321,antigua & barbuda\n323,bahamas\n324,barbados\n327,cuba\n328,dominica\n329,dominican republic\n330,grenada\n332,haiti\n333,jamaica\n339,st. lucia\n340,st. vincent & the grenadines\n341,trinidad & tobago\n343,west indies\n344,\"caribbean, not specified\"\n360,argentina\n361,bolivia\n362,brazil\n363,chile\n364,colombia\n365,ecuador\n368,guyana\n369,paraguay\n370,peru\n372,uruguay\n373,venezuela\n374,south america\n399,\"americas, not specified\"\n400,algeria\n407,cameroon\n408,cabo verde\n412,congo\n414,egypt\n416,ethiopia\n417,eritrea\n420,gambia\n421,ghana\n423,guinea\n427,kenya\n429,liberia\n430,libya\n436,morocco\n440,nigeria\n444,senegal\n447,sierra leone\n448,somalia\n449,south africa\n451,sudan\n453,tanzania\n454,togo\n457,uganda\n459,democratic republic of congo (zaire)\n460,zambia\n461,zimbabwe\n462,africa\n463,\"eastern africa, not specified\"\n464,\"northern africa, not specified\"\n467,\"western africa, not specified\"\n468,\"other africa, not specified\"\n501,australia\n508,fiji\n511,marshall islands\n512,micronesia\n515,new zealand\n523,tonga\n527,samoa\n554,\"other us island areas, oceania, not specified, or at sea\"\n"
country<- read_csv(countries) %>% rename(POBP=code)
country$Country<- tools::toTitleCase(country$Country)
world_map_countries<- inner_join(country,country_df,by="Country")
pob_of_immigrants<- non_citizen %>%
inner_join(country,by="POBP") %>%
group_by(Country) %>%
summarise(Population=n()) %>%
arrange(desc(Population))
pob_of_immigrants## # A tibble: 159 x 2
## Country Population
## <chr> <int>
## 1 Mexico 305425
## 2 India 51350
## 3 China 46473
## 4 El Salvador 30963
## 5 Philippines 27679
## 6 Canada 23167
## 7 Guatemala 22537
## 8 Korea 18070
## 9 Cuba 17813
## 10 Dominican Republic 16325
## # … with 149 more rowspob_of_immigrants<- inner_join(world_map_countries,pob_of_immigrants)
pob_of_immigrants<- pob_of_immigrants %>% arrange(desc(Population))
l1 <- list(color = toRGB("grey"), width = 0.5)
# specify map projection/options
g1 <- list(
showframe = FALSE,
showcoastlines = FALSE,
projection = list(type = 'Mercator')
)
plot_geo(pob_of_immigrants) %>%
add_trace(
z = ~Population, color = ~Population, colors = 'Reds',
text = ~Country, locations = ~CODE, marker = list(line = l)
) %>%
colorbar(title = 'Population of Immigrants') %>%
layout(
title = "Number of Immigrants in U.S from each Country",
geo = g1
)
3) More females migrates from China, Phillipines and Korea.
#number of male and female from each country.
gender_of_immigrants<- states_noncitizen %>%
inner_join(country,by="POBP") %>%
filter(!Country=="Mexico") %>%
mutate(Gender=case_when(SEX == 1 ~ "Males",
SEX == 2 ~ "Females")) %>%
group_by(Country,Gender) %>%
summarise(n=n()) %>%
arrange(desc(n)) %>%
rename(Count=n) %>%
head(20)
gender_of_immigrants## # A tibble: 20 x 3
## # Groups: Country [11]
## Country Gender Count
## <chr> <chr> <int>
## 1 India Males 27259
## 2 China Females 24575
## 3 India Females 24091
## 4 China Males 21898
## 5 Philippines Females 16514
## 6 El Salvador Males 16097
## 7 El Salvador Females 14866
## 8 Guatemala Males 13438
## 9 Canada Females 12394
## 10 Philippines Males 11165
## 11 Canada Males 10773
## 12 Korea Females 9942
## 13 Cuba Males 9658
## 14 Guatemala Females 9099
## 15 Dominican Republic Females 8880
## 16 Cuba Females 8155
## 17 Korea Males 8128
## 18 Honduras Males 7903
## 19 Dominican Republic Males 7445
## 20 Vietnam Females 7286#plotting.
ggplot(gender_of_immigrants, aes(x=fct_reorder(Country,desc(Count)),Count, color=Gender)) +
geom_point() +
geom_segment( aes(x=Country, xend=Country, y=0, yend=Count)) +
theme(axis.text.x = element_text(angle = 90)) +
xlab("Country") + ylab("Population") + labs(title = "Gender of Immigrants from 20 countries.")
In this code, I removed the number of Mexican immigrants because there was a huge difference and the plot was not easily interpretable. From the above plot we can see, that number of females are more in the countries China,Phillipines,Korea and Dominican Republic. An interesting thing to notice is that, from Honduras, there is unsimilarity in the number of males and females. There are less than 7k females from Honduras that migrate to U.s. And the same is for the males from Vietnam.
4)District of Columbia(DC) and Connecticut(CT) are the states with highest Per Capita Income of immigrants.
#calculating Per capita income of immigrants in each state.
statewise_income<- states_noncitizen %>%
filter(!is.na(PINCP)) %>%
group_by(abbr) %>%
summarise(count=n(),Per_capita_income=sum(as.numeric(PINCP))/count) %>%
arrange(desc(Per_capita_income)) %>%
rename(state=abbr)
statewise_income## # A tibble: 51 x 3
## state count Per_capita_income
## <chr> <int> <dbl>
## 1 DC 2158 49081.
## 2 CT 9423 41032.
## 3 NH 1372 37829.
## 4 NJ 32931 36459.
## 5 WA 19134 35824.
## 6 MA 20478 35819.
## 7 MD 15936 33876.
## 8 DE 1571 33497.
## 9 VA 17350 33478.
## 10 VT 471 32652.
## # … with 41 more rowsstatewise_income<-inner_join(statewise_income,states_df,by=c("state"="code")) %>% select(state,Per_capita_income)
#plotting on the U.S map
statewise_income$hover <- with(statewise_income, paste(state, '<br>', "Income", Per_capita_income))
l3 <- list(color = toRGB("white"), width = 2)
g3 <- list(
scope = 'usa',
projection = list(type = 'albers usa'),
showlakes = TRUE,
lakecolor = toRGB('white')
)
plot_geo(statewise_income, locationmode = 'USA-states') %>%
add_trace(
z = ~Per_capita_income, text = ~hover, locations = ~state,
color = ~Per_capita_income, colors = 'Reds'
) %>%
colorbar(title = "Income") %>%
layout(
title = 'Total Income of immigrants in each state',
geo = g3
)
5) California has most amount of immigrants, but it has the most amount of Jobless Immigrants too!
#jobless immigrants in each state.
jobless_immigrants <- non_citizen %>%
inner_join(statepop,by="ST") %>%
filter(ESR==3) %>%
group_by(abbr) %>%
summarise(jobless=n()) %>%
arrange(desc(jobless)) %>%
rename(state=abbr)
jobless_immigrants## # A tibble: 51 x 2
## state jobless
## <chr> <int>
## 1 CA 10373
## 2 TX 3581
## 3 NY 3554
## 4 FL 3279
## 5 NJ 1482
## 6 IL 1297
## 7 MA 961
## 8 AZ 825
## 9 WA 782
## 10 GA 723
## # … with 41 more rows#plotting it on a static U.S map
plot_usmap(data = jobless_immigrants, values = "jobless", color = "black") +
scale_fill_continuous(name = "Jobless Immigrants", label = scales::comma) +
theme(legend.position = "right")
In this U.S map, it can be seen that California(CA) has the most amount of Jobless Immigrants. I defined people as Jobless by using “ESR==3” (Employment Status Recode). The value “3” means Unemloyed.. Texas also has a large amount of jobless people( around 3.5k).The least amount of jobless immigrants are in Alaska(AK), but that is because the total number of immigrants are very less too.
6) People from China migrate the most to U.S to pursue a Bachelor’s or higher level degree.
#immigrants currently pursuing bachelor's or higher level degree from U.S.
immigrants_studying<- non_citizen %>%
inner_join(country,by="POBP") %>%
filter(SCHG>=15) %>%
group_by(Country) %>%
summarise(Population=n()) %>%
arrange(desc(Population))
immigrants_studying## # A tibble: 159 x 2
## Country Population
## <chr> <int>
## 1 China 12220
## 2 Mexico 10701
## 3 India 5228
## 4 Korea 3946
## 5 Philippines 2186
## 6 Canada 1874
## 7 Vietnam 1857
## 8 Brazil 1558
## 9 Saudi Arabia 1371
## 10 Colombia 1311
## # … with 149 more rowsimmigrants_studying<- inner_join(world_map_countries,immigrants_studying) ## Joining, by = "Country"immigrants_studying<- immigrants_studying %>% arrange(desc(Population)) %>% select(-GDP..BILLIONS.)
l4 <- list(color = toRGB("grey"), width = 0.5)
# specify map projection/options
g4 <- list(
showframe = FALSE,
showcoastlines = FALSE,
projection = list(type = 'Mercator')
)
plot_geo(immigrants_studying) %>%
add_trace(
z = ~Population, color = ~Population, colors = 'Purples',
text = ~Country, locations = ~CODE, marker = list(line = l)
) %>%
colorbar(title = 'Population of Immigrants') %>%
layout(
title = "Countrywise immigrants pursuing Bachelor's or higher level degree in U.S",
geo = g4
)The most number of immigrants that migrate to U.S are from Mexico. So it can be said that most number of students must also be from Mexico. But it’s not true. China has the highest number of students that migrate to U.S.
7)People who have a degree in Mettalurgical Engineering have the highest income.
#top field of degree based on income.
fodwise_income<-states_noncitizen %>%
filter(!is.na(FOD1P)) %>%
group_by(FOD1P) %>%
summarise(Number_of_People=n(), per_capita_income= sum(as.numeric(PINCP))/Number_of_People) %>%
arrange(desc(per_capita_income)) %>%
head(10)
#plotting the data.
ggplot(fodwise_income,aes(fct_reorder(FOD1P,desc(per_capita_income)),per_capita_income, fill=per_capita_income)) +
geom_bar(stat = "identity") +
theme(axis.text.x = element_text(angle = 90,hjust = 1)) +
xlab("Field of degree") +
ylab("Income in millions") +
scale_x_discrete(labels=c("Metallurgical","Maths & Comp. Science","Petroleum","Actuarial Science","Geosiences","Physiology","Health Programs","Geological","Astronomy","Military Tech")) +
labs(title="Top 10 field of degrees based on income",fill="Income")
The above table shows that top 10 fields of degree based on the income by that degree holders. Metallurgical Engineering is field with highest income.
8) Indian students are the students who have a Degree and are looking for a job in U.S.
#nationality wise top 5: immigrants who have a Bachelor's or higher level degree and are looking for a job
jobless_degreeholder_immigrants <- states_noncitizen %>%
inner_join(country,by="POBP") %>%
filter(SCHL>=21 & NWLK==1) %>%
group_by(Country) %>%
summarise(Number_of_Immigrants=n()) %>%
arrange(desc(Number_of_Immigrants)) %>%
head(5)
#plotting
ggplot(jobless_degreeholder_immigrants,aes(fct_reorder(Country,desc(Number_of_Immigrants)),Number_of_Immigrants,fill=Country)) +
geom_bar(stat="identity") + xlab("Country") + ylab("Immigrants") + labs(title="Number of Jobless Degree Holders")
The most number of jobless degree holders immigrants are from India. The number is not very large as compared to total immigrants from India.
- Australian immigrants have the highest income in U.S compared to other immigrants.
#income of immigrants.
pob_income<- non_citizen %>%
filter(!is.na(PINCP)) %>%
inner_join(country,by="POBP") %>%
group_by(Country) %>%
summarise(count=n(),Per_capita_income=sum(as.numeric(PINCP))/count) %>%
arrange(desc(Per_capita_income))
pob_income <- inner_join(world_map_countries,pob_income) %>% select(-GDP..BILLIONS.) %>% arrange(desc(Per_capita_income)) %>% select(-c(POBP,count))## Joining, by = "Country"pob_income## # A tibble: 128 x 3
## Country CODE Per_capita_income
## <chr> <fct> <dbl>
## 1 Australia AUS 77700.
## 2 Belgium BEL 71384.
## 3 Denmark DNK 70552.
## 4 Switzerland CHE 69355.
## 5 Ireland IRL 69223.
## 6 France FRA 66881.
## 7 New Zealand NZL 66485.
## 8 Norway NOR 66087.
## 9 Netherlands NLD 65353.
## 10 Cyprus CYP 64679.
## # … with 118 more rows#plotting the data on world map.
l2 <- list(color = toRGB("grey"), width = 0.5)
g2 <- list(
showframe = FALSE,
showcoastlines = FALSE,
projection = list(type = 'Mercator')
)
plot_geo(pob_income) %>%
add_trace(
z = ~Per_capita_income, color = ~Per_capita_income, colors = 'Greens',
text = ~Country, locations = ~CODE, marker = list(line = l)
) %>%
colorbar(title = 'Income') %>%
layout(
title = "Countrywise Per Capita Income of Immigrants",
geo = g2
)10) The females with a degree in Metallurgical Engineering earn a lot less than males.
options(scipen = 999)
#income of males with a degree in Metallurgical Engineering.
male_income<- non_citizen %>%
filter(!is.na(PINCP) & SEX==1 & FOD1P==2415) %>%
select(PINCP,SEX,FOD1P)
#income of females with a degree in Metallurgical Engineering.
female_income<- non_citizen %>%
filter(!is.na(PINCP) & SEX==2 & FOD1P==2415) %>%
select(PINCP,SEX,FOD1P)
#testing the incomes of males and females.
t.test(male_income$PINCP,female_income$PINCP)##
## Welch Two Sample t-test
##
## data: male_income$PINCP and female_income$PINCP
## t = 6.013, df = 55.718, p-value = 0.0000001471
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
## 51869.41 103704.50
## sample estimates:
## mean of x mean of y
## 119888.01 42101.06
I ran a t-test to check the means of incomes of Females and Males with Metallurgical engineering. As the variance of both of them were different, therefore a Welch Test was run by default.The p-value that I got is way lower than the significant level(0.05). Therefore the null hypothesis that the means are equal, is rejected.
Dicussion:
The main aim of my project was to do an exploratory analysis in immigrants in U.S. So, I selected the colummns that I felt would contribute to find something interesting about the trends of immigrants in U.S. The first thing I found was that most of the immigrants from all around the migrated to California state. And the least went to Montana state.
Next, Mexicans migrate the most to U.S and second to Mexicans are the immigrants from India .
I checked the gender of immigrants from all the countries and there were some interesting thihngs about the data.I found out that females from China, Canada, Korea , Phillipines and some other countries migrate to U.S more than the males of the country.
Although the number of immigrants are the highest in California, still the per capita income of immigrants is the highest in District of Columbia(DC).
After looking at the income, I thought I should look that people with which degree earn the most. So I found out that Metallurgical Engineering degree holders have the highest income in United States of America.
Then I found out that the highest amount of jobess imimigrants are in California state.
I wanted to check that from which country, most students go to U.S to pursue a Bachelor’s or a higher level degree. I found a very interesting thing that even though the total number of immigrants in U.S were the highest from Mexico, still the highest number of students immigrated from India. Then I found out that, Indians are the one’s who have a degree and are still finding a job in U.S. This doesn’t necessarily means that they pursued the degree from U.S itself. There may be a possibility that the students have completed their degree from India and are looking for job opportunities in U.S.
Then as I went on the exploratory analysis, I thought of finding immigrants from which country earn the most in U.s. Australians are the one’s who earn the most. Except them, immigrants from europe also earn a lot, as compared to immigrants from other countries.
When I was finding the highest earning field of degree,a thought came to my mind that I should check whether there is some bias in the income based on the gender of a person.And I found out that males earn more than double the amount females earn.
I am quite confident that the analysis that I have done is accurate and I believe most of my conclusions.Although,I am not confident enough in my linear models, because the results were not as expected by me.
The limitations in my project is that I couldn’t predict the income of immigrants based on various variables.
Appendix:
1) Predicting income based on various variables:
#filtering graduates from immigrants data.
grads<- states_noncitizen %>%
filter(SCHL>=21) %>%
group_by(abbr)
#income of grads in each state.
grads_income<- grads %>% filter(!is.na(PINCP)) %>% summarise(number_of_grads=n(),literacy=sum(as.numeric(SCHL)),per_capita_income = sum(as.numeric(PINCP))/number_of_grads)%>% arrange(desc(literacy)) %>% rename(state=abbr)
grads_income## # A tibble: 51 x 4
## state number_of_grads literacy per_capita_income
## <chr> <int> <dbl> <dbl>
## 1 CA 44357 958606 62509.
## 2 TX 19111 413432 55622.
## 3 NY 18678 404367 63256.
## 4 FL 15828 341306 47209.
## 5 NJ 11595 250611 65470.
## 6 IL 8129 176138 53228.
## 7 MA 7672 168507 62769.
## 8 VA 6325 137175 56710.
## 9 WA 5739 124265 70304.
## 10 MD 5454 119295 56700.
## # … with 41 more rows
Calculating literacy rates in each state.
#literacy in each state
literacy_in_state<- grads_income %>% select(state,literacy)
literacy_in_state## # A tibble: 51 x 2
## state literacy
## <chr> <dbl>
## 1 CA 958606
## 2 TX 413432
## 3 NY 404367
## 4 FL 341306
## 5 NJ 250611
## 6 IL 176138
## 7 MA 168507
## 8 VA 137175
## 9 WA 124265
## 10 MD 119295
## # … with 41 more rows
Predicting income using literacy.
income_literacy_lm <- lm(per_capita_income~literacy,grads_income)
summary(income_literacy_lm)##
## Call:
## lm(formula = per_capita_income ~ literacy, data = grads_income)
##
## Residuals:
## Min 1Q Median 3Q Max
## -12398.1 -7266.6 -357.5 4803.2 26541.4
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 48903.165415 1375.964344 35.541 < 0.0000000000000002 ***
## literacy 0.023181 0.007737 2.996 0.00428 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 8640 on 49 degrees of freedom
## Multiple R-squared: 0.1548, Adjusted R-squared: 0.1376
## F-statistic: 8.976 on 1 and 49 DF, p-value: 0.004283
The adjusted r-squared is just 0.1376. This means that the model is just 13% confident.
income_numberofgrads_lm<- lm(per_capita_income~number_of_grads,grads_income)
summary(income_numberofgrads_lm)##
## Call:
## lm(formula = per_capita_income ~ number_of_grads, data = grads_income)
##
## Residuals:
## Min 1Q Median 3Q Max
## -12408.8 -7275.2 -363.5 4823.0 26543.1
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 48914.3409 1375.8687 35.552 < 0.0000000000000002 ***
## number_of_grads 0.4996 0.1673 2.985 0.00441 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 8645 on 49 degrees of freedom
## Multiple R-squared: 0.1539, Adjusted R-squared: 0.1366
## F-statistic: 8.913 on 1 and 49 DF, p-value: 0.00441
While predicting using the number of graduates , the model was still just 13% confident.
income_literacy_grads_lm<- lm(per_capita_income~literacy+number_of_grads,grads_income)
summary(income_literacy_grads_lm)##
## Call:
## lm(formula = per_capita_income ~ literacy + number_of_grads,
## data = grads_income)
##
## Residuals:
## Min 1Q Median 3Q Max
## -10769 -5585 -1267 3275 24630
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 47164.109 1370.312 34.419 < 0.0000000000000002 ***
## literacy 7.600 2.346 3.239 0.00218 **
## number_of_grads -163.788 50.718 -3.229 0.00224 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 7912 on 48 degrees of freedom
## Multiple R-squared: 0.3057, Adjusted R-squared: 0.2767
## F-statistic: 10.57 on 2 and 48 DF, p-value: 0.0001576
When predicting the income using two variables, i.e. literacy and total number of grads in a state, the model was 27% confident.
This code gave something interesting results, because I didn’t think that a country would take immigrants in their armed forces.This is initeresting, but irrevelant to my project.
#number of people in armed forces in each state ,immigrants in armed forces?? \(0_o)/
statewise_armed_forces<- states_noncitizen %>%
group_by(abbr) %>%
summarise(armed_forces=n()) %>%
arrange(desc(armed_forces)) %>%
rename(state=abbr)
statewise_armed_forces## # A tibble: 51 x 2
## state armed_forces
## <chr> <int>
## 1 CA 225964
## 2 TX 113504
## 3 NY 76376
## 4 FL 72484
## 5 NJ 35270
## 6 IL 32593
## 7 AZ 22053
## 8 GA 21944
## 9 MA 21848
## 10 WA 20448
## # … with 41 more rows